effects of vegetation cover, presence of a native ant species, and human disturbance on colonization...

Contributed Paper

Effects of Vegetation Cover, Presence of a Native AntSpecies, and Human Disturbance on Colonizationby Argentine AntsKATHERINE FITZGERALD∗ AND DEBORAH M. GORDONDepartment of Biology, Stanford University, 371 Serra Mall, Stanford, CA 94305-5020, U.S.A.

Abstract: The spread of non-native invasive species is affected by human activity, vegetation cover, weather,and interaction with native species. We analyzed data from a 17-year study of the distribution of the non-native Argentine ant (Linepithema humile) and the native winter ant (Prenolepis imparis) in a preserve innorthern California (U.S.A.). We conducted logistic regressions and used model selection to determine whetherthe following variables were associated with changes in the distribution of each species: presence of conspecificsat neighboring sites, distance to development (e.g., roads, buildings, and landscaped areas), proportion of veg-etation cover taller than 0.75 m, elevation, distance to water, presence of both species at a site, temperature,and rainfall. Argentine ants colonized unoccupied sites from neighboring sites, but the probability of appear-ance and persistence decreased as distance to development, vegetation cover, and elevation increased. Winterants appeared and persisted in sites with relatively high vegetation cover (i.e., highly shaded sites). Presence ofthe 2 species was negatively associated in sites with high vegetation cover (more winter ants) and sites neardevelopment (more Argentine ants). Probability of colonization of Argentine ants decreased where winterants were most persistent. At sites near development within the preserve, abundant Argentine ant populationsmay be excluding winter ants. The high abundance of Argentine ants at these sites may be due to immigrationfrom suburban areas outside the preserve, which are high-quality habitat for Argentine ants. In the interiorof the preserve, distance from development, low-quality habitat, and interaction with winter ants may incombination exclude Argentine ants. Interactions among the variables we examined were associated withlow probabilities of Argentine ant colonization in the preserve.

Keywords: edge effect, Prenolepis imparis, propagule pressure, vegetation cover

Efectos de la Cobertura Vegetal, Presencia de Especies de Hormigas Nativas y Perturbacion Humana sobre laColonizacion por Hormigas Argentinas

Resumen: La expansion de especies invasoras no nativas es afectada por la actividad humana, la coberturavegetal, el tiempo y la interaccion con especies nativas. Analizamos datos de un estudio de 17 anos de ladistribucion de la hormiga argentina no nativa (Linepithema humile) y la hormiga nativa (Prenolepis imparis)en una reserva en el norte de California (E.U.A.). Realizamos regresiones logısticas y utilizamos seleccion demodelos para determinar si las siguientes variables se asociaban con cambios en la distribucion de cadaespecie: presencia de individuos de la misma especie en los alrededores, distancia a areas desarrolladas(e. g., caminos, edificios y jardines), proporcion de cobertura vegetal mayor a 0.75 m, altura, distancia alagua, presencia de ambas especies, temperatura y precipitacion. Las hormigas argentinas colonizaron sitiosno ocupados de sitios adyacentes, pero la probabilidad de aparicion y persistencia decrecio a medida queincremento la distancia al desarrollo, la cobertura vegetal y la altitud. Prenolepis imparis aparecio y persistioen sitios con cobertura vegetal relativamente amplia (i e., sitios muy sombreados). La presencia de las 2

∗email [email protected] submitted December 23, 2010; revised manuscript accepted October 5, 2011.

525Conservation Biology, Volume 26, No. 3, 525–538C©2012 Society for Conservation BiologyDOI: 10.1111/j.1523-1739.2012.01836.x

526 Argentine Ant Colonization Probability

especies se asocio negativamente en sitios con amplia cobertura vegetal (menos P. imparis) y en sitios condesarrollo (mas L. humile). La probabilidad de colonizacion de L. humile disminuyo donde P. imparis eran maspersistentes. En sitios cercanos al desarrollo, las abundantes poblaciones de L. humile pueden estar excluyendoa P. imparis. Al gran abundancia de L. humile en estos sitios se puede deber a la inmigracion desde areassuburbanas fuera de la reserva, que son habitat de buena calidad para L. humile. En el interior de la reserva,la distancia a areas desarrolladas, habitat de baja calidad y la interaccion con P. imparis combinadas puedeexcluir a L. humile. Las interacciones entre las variables que examinamos estuvieron asociadas con bajaprobabilidad de colonizacion de la reserva por L. humile.

Palabras Clave: cobertura vegetal, efecto de borde, Prenolepis imparis, presion de propagulos

Introduction

Non-native invasive species reduce probabilities of persis-tence of native species in many communities and ecosys-tems (Sakai et al. 2001). Non-native ants, for example,disrupt ant communities and interspecific interactionsin the areas they colonize (Holway et al. 2002; O’Dowdet al. 2003). The geographic spread of non-native inva-sive species and their effects on native species are associ-ated with environmental covariates including vegetationcover, weather and climate, and human activity (Thomas& Holway 2005; Going et al. 2009; Kestrup & Ricciardi2009). Human land use can facilitate invasive species’colonization of previously uninvasible areas (Menke &Holway 2006; Pauchard et al. 2009) or create con-ditions that allow invaders to exclude native species(MacDougall & Turkington 2005; Menke et al. 2007; King& Tschinkel 2008).

In a preserve in northern California (U.S.A.), we exam-ined the effects of proximity to developed areas, weather,distance to water, elevation, proportion of vegetationcover taller than 0.75 m, and presence of a native antspecies on site-level colonization by the Argentine ant(Linepithema humile) . The Argentine ant has invadedareas worldwide, especially in Mediterranean climatezones. Its global spread is associated with human activity,and in many invaded regions it inhabits primarily areasaround human habitations (Suarez et al. 2001; Carpinteroet al. 2003; Rowles & Silverman 2009). In arid regions,heat and low humidity limit spread (Menke & Holway2006; Schilman et al. 2007), and Argentine ants have col-onized natural areas only along riparian corridors or ur-ban edges that receive water runoff (Suarez et al. 1998;Holway 2005; Holway & Suarez 2006). In more mesicregions, such as some parts of Hawaii, Argentine antshave colonized natural areas more extensively (Krushel-nycky et al. 2005; Hartley et al. 2010) than in arid regionsand spread is limited by cold temperatures, which hin-der colony growth and reproduction (Hartley & Lester2003; Abril et al. 2008). Argentine ants’ moisture andtemperature requirements may preclude colonization ofuplands in some regions (Ward 1987) and forests in oth-ers (Krushelnycky et al. 2005; Ward & Harris 2005).

Where Argentine ants have colonized, few native antspecies coexist with them (Holway et al. 2002; Carpin-

tero et al. 2007; Rowles & O’Dowd 2007). In California,however, the native winter ant (Prenolepis imparis) sur-vives in areas colonized by Argentine ants (Ward 1987;Suarez et al. 1998; K. Fitzgerald, T. Sorrells, and D.M.Gordon, unpublished). Both species feed on honeydewproduced by phloem-sucking insects (Nygard et al. 2008;Rowles & Silverman 2009). They coexist partly becausetheir nesting and seasonal ecology differ. Argentine antsmove nest locations frequently (Heller & Gordon 2006)and are most active in September (Suarez et al. 1998;Sanders et al. 2001). In contrast, winter ants live in deep,long-used nests and are not known to migrate to newnests (Tschinkel 1987). Their activity peaks during lateautumn and early spring, and they often estivate duringthe summer, but specific timing of activities varies by lo-cation (Tschinkel 1987; Suarez et al. 1998; K. Fitzgerald,T. Sorrells, and D.M. Gordon, unpublished).

From 1993 through 2009, twice-yearly surveys of antdistributions were conducted at Jasper Ridge BiologicalPreserve (Santa Clara County, California), a 481-ha naturalarea owned by Stanford University, to track the coloniza-tion of the preserve by Argentine ants and the response ofnative ants to the colonization. When the survey began in1993, Argentine ants already occupied many peripheral,low-elevation areas of the preserve (Human et al. 1998).The source of the ants in Jasper Ridge was suburbanareas near the preserve. Between 1993 and 2001, Argen-tine ants progressively colonized the preserve in seasonalpulses. The invasion front progressed to new sites dur-ing summer and retreated during winter. In spring, theArgentine ants occupied fewer sites than the previous au-tumn (Sanders et al. 2001). These pulses reflected eachcolony’s yearly cycle of expansion into a network of manysmall, ephemeral, and widespread nests in summer andtheir contraction into larger clusters of relatively perma-nent nests in winter (Heller & Gordon 2006).

Since 2001, few hectares of Jasper Ridge have beencolonized for the first time, and summer expansions havebeen offset by winter contractions. Here, we examinedthe environmental variables associated with changes indistributions of Argentine and winter ants (P. imparis)since 1993. Taking into account the short dispersal dis-tance of the Argentine ant, we investigated whetherchanges in Argentine ant distribution were associatedwith proximity to development (e.g., roads, buildings,

Conservation BiologyVolume 26, No. 3, 2012

Fitzgerald & Gordon 527

and landscaped areas); environmental variables (e.g., veg-etation cover and elevation); presence of winter ants;and weather. We then investigated whether these vari-ables and Argentine ant presence are related to changesin distributions of winter ants.

Methods

Ant Surveys

Jasper Ridge Biological Preserve (37◦24′29′′N,122◦13′39′′W) is bordered by low-density suburbandevelopment, including buildings, farms, and roads; andthe property itself contains a few buildings, parkinglots, paved and dirt roads, and hiking and equestriantrails. It includes sections of San Francisquito Creekand its tributaries, including Searsville Lake and Dam.The preserve contains several different vegetationcommunities, including serpentine soil-based grasslanddominated by native plant species, annual grasslanddominated by non-native species, chaparral, coyotebrush (Baccharis pilularis), and poison oak (Toxicoden-dron diversilobum) scrub, oak woodland, evergreenwoodland, riparian woodland, and wetland.

We surveyed for ants twice yearly, in May and Septem-ber, from May 1993 through September 2009. Surveypoints were arranged on a 100-m grid superimposed overthe accessible areas (approximately 69%) of the preserve.We surveyed 334 points at least once and 121 points dur-ing all 34 surveys (mean of 25 [SD 11.6] surveys perpoint). In each survey, a 20-m radius around each pointwas searched once for 5 min. We identified all ants togenus. Through 1995 points with no ants were baitedwith honey traps for 24 hours (Sanders et al. 2001).Through 1996 ants were sometimes identified only asArgentine or native ants. For more information on surveymethods see Human et al. (1998), Sanders et al. (2001),and Heller et al. (2008).

For each survey point, we used the Santa Clara CountyLIDAR (light detection and ranging) data set, which pro-vides measures of ground elevation and vegetation heighttaken from an aircraft, to calculate the average elevationof the 1 ha surrounding the point and the proportion ofthe hectare covered by trees and shrubs (i.e., vegetationtaller than 0.75 m). We used maps of Jasper Ridge andan aerial photograph to calculate the distance in metersfrom each survey point to the nearest water body and tothe nearest developed area. Developed areas, clusteredwithin the western portion of the preserve and along itsboundary, included 7 buildings (field station, ranger res-idence, an art studio, an occasionally used horse stable,and several storage buildings), 3 gravel parking lots, trails,frequently traveled paved and gravel roads within the pre-serve, and the preserve boundary. We performed thesecalculations in ArcGIS (version 9.2., ESRI, Redlands, Cali-fornia) (additional details in Supporting Information).

Weather Data

We obtained data on precipitation and average minimumand maximum temperatures for each month betweenNovember 1992 and September 2009 from the PRISMClimate Group (2010). We averaged data from the 2 (4km × 4 km) PRISM grid squares that overlap Jasper Ridge(centered at 37◦24′59′′N, 122◦15′W and 37◦24′59′′N,122◦12′31′′W). Following Heller et al. (2008), we calcu-lated total rainfall and average minimum and maximumtemperatures for each winter (November–April) and sum-mer (May–September).

Analyses

Argentine ants spread with human assistance or on theirown by budding, when a queen and workers walk to anew nest location (Suarez et al. 2001; Ingram & Gor-don 2003). At Jasper Ridge, spread through buddingpredominates, with a maximum rate of 100–150 m/year(Suarez et al. 2001; DiGirolamo & Fox 2006), and longer-distance dispersal is limited to males during mating flights(Ingram & Gordon 2003). Because our survey sites wereon a 100-m grid, a site was likely to be colonized onlyif Argentine ants were already present at 1 of the 8neighboring sites, which were either 100 or 141 maway.

In analyses of species’ invasions in which data are col-lected from sites arrayed in a grid, spatial autocorrela-tion may lead to spurious associations between variablesassociated with colonization and features of sites nearother sites that are already colonized because probabilityof invasion may be associated with distance from con-specifics (Bini et al. 2009). We addressed autocorrelationby explicitly including the presence of Argentine antsat neighboring sites in our analyses. One of our fixed-effects variables was the proportion of surveyed neigh-boring sites at which Argentine ants were observed. Wealso examined whether environmental variables were as-sociated with probability of persistence of Argentine antsat sites they had already colonized.

We used model selection to determine whether vari-ables were associated with the appearance by Argentineants at previously unoccupied sites and with Argentineants’ persistence at sites. We examined the changes inthe presence of Argentine ants from each survey periodto the next and from each survey period to the one in thesame season the following year. We evaluated the weightof evidence for alternative models of 8 probabilities oftransition, 4 from absence to presence of Argentine antsand 4 from presence to absence of Argentine ants. Weexamined probabilities of each of those 2 types of transi-tion from spring to autumn in a given year (n = 17 years),from autumn of a given year to spring of the followingyear (n = 16), from autumn of a given year to autumnof the following year (n = 16), and from spring of agiven year to spring of the following year (n = 16) . In



each case, the binomial response variable was the pres-ence or absence of Argentine ants in the second surveyperiod.

We fit logistic regressions in R (version 2.10) (R De-velopment Core Team, Vienna, Austria) with generalizedlinear mixed models (Bates & Maechler 2009). Modelsincluded random effects of site and year to account forrepeated observations at the same sites. The fixed-effectsvariables were the proportion of surveyed neighboringsites in which Argentine ants were present during thefirst survey period in each comparison, the presence ofwinter ants at the site during the first survey period ineach comparison, distance to development, proportionof the 1 ha surrounding the site covered with vegeta-tion taller than 0.75 m, elevation, distance to water, rain-fall, temperature, and the interactions among variables.Weather variables were included for the time betweenthe 2 survey periods. Each model computed the proba-bility (p) of Argentine ant presence at the next surveyperiod, given the site and weather variables, with a logis-tic function, p = 1/(1 + e−f (X)), where f (X) represents alinear combination of the fixed and random effects.

To select the variables most strongly associated withappearance or persistence, we used forward selectionwith backward elimination and the Akaike informationcriterion (AIC) as the selection criterion (Burnham &Anderson 1998). This procedure selects the covariatesmost strongly associated with the response variables. Westopped adding variables when no new variables reducedthe AIC by more than 2 because models with �AIC < 2are generally considered equivalent (Burnham & Ander-son 1998). When no new variables could be added, weadded 2-way interactions for the selected variables (sameprocedure as for single variables). If, in a given step, thedifference in AIC values of 2 models was <2, we con-tinued model selection with both models in parallel. Wewere able to identify a single best model for each of the 8probabilities of transition and did not perform model av-eraging. Because we used forward selection rather thanexploring the entire model space, we did not calculateabsolute �AIC.

To elucidate the effects of continuous variables and ofinteractions between winter ant presence and continu-ous variables on the transition probabilities, we graphedcurves of the fitted models for each transition. We plot-ted the effect of each term in turn and held all othervariables constant at their mean values. We computedthe value of each variable at which the probability ofappearance or persistence, as calculated by the logisticcurve, was halfway between its minimum and maximumvalues.

We conducted the same analyses for winter ants aswe did for Argentine ants, except that the first fixed-effect variable was the proportion of surveyed neighbor-ing sites in which winter ants were present and the sec-ond was the presence of Argentine ants. We discarded

data points when native ants at a site were not iden-tified to species for either of the survey periods beingcompared.

Because Argentine ants and winter ants differ in sea-sonal activity, observations of the 2 species requireslightly different interpretations. Argentine ants are ac-tive throughout the day and year (Human et al. 1998),so differences in occupancy from one survey period tothe other probably reflect changes in local density of Ar-gentine ants. In contrast, when winter ants were not ob-served during a survey period, especially in the autumn,they may have been present but estivating, so our datamay underestimate their distributions. The data show thesites in which winter ants were present aboveground inthe spring and autumn, the periods when encounters be-tween the 2 species are most likely (K. F., T. Sorrells, andD. G., unpublished).

Results

Presence of Conspecifics at Neighboring Sites

Argentine ants were more likely to appear and persist atsites adjacent to those with Argentine ants (Figs. 1a & c).The proportion of neighboring sites with Argentine antswas the covariate most strongly associated with appear-ance and persistence and was the first or second selectedvariable for all 8 models (Table 1).

For winter ants, conspecific neighbors were less con-sistently associated with appearance and persistence thanfor Argentine ants. This variable was selected in 3 of 8models. In the spring, winter ants appeared and persistedmore often at sites that had many winter ant neighborsthe previous spring or autumn (Table 2) than at sites withfew conspecific neighbors.

Distance from Development

Argentine ants appeared and persisted most often at sitesclosest to developed areas (Figs. 1b & d). After propor-tion of neighboring sites with Argentine ants, proximityto development was the covariate most strongly associ-ated with Argentine ant appearance and persistence andwas the first or second variable selected in 7 of the 8 mod-els (Table 1). From one autumn to the next, there was aweak interaction, selected last of 9 covariates, betweendistance to development and winter minimum temper-ature. As winter minimum temperatures decreased, theassociation between distance to development and persis-tence increased.

Distance to development was not associated with win-ter ant appearance or persistence. Proximity to devel-oped areas was associated with appearance or persis-tence of winter ants only when Argentine ants werepresent. The negative association of Argentine ants withappearance at sites by winter ants decreased as distancefrom developed areas increased (Fig. 2a).



Figure 1. Seasonal patterns in effects of presence of Argentine ants in 8 neighboring sites on a grid and developedareas (buildings, roads, and landscaped areas) on probability of appearance and persistence of Argentine ants ata site: appearance of Argentine ants at previously unoccupied sites (a) as a function of proportion of 8neighboring sites with Argentine ants, (b) as a function of distance to nearest developed area and persistence ofArgentine ants, (c) as a function of proportion of 8 neighboring sites with Argentine ants, and (d) as a function ofdistance to nearest developed area (heavy solid lines, from one autumn to the next; light solid lines, from spring toautumn; heavy dashed lines, from one spring to the next; light dashed lines, from autumn to spring). Probabilitieswere calculated with the logistic functions computed for selected models (Table 1), and we assumed values of allvariables were at their mean except the variable shown on the x-axis.

Vegetation Cover

As vegetation cover decreased, the probability of Argen-tine ant persistence increased. Logistic curves reachedhalf their maxima when the proportion of cover tallerthan 0.75 m was 0.64–0.79, depending on season(Fig. 3a). Vegetation cover was moderately associatedwith persistence, selected second of 7, third of 5, fifth of7, and third of 9 terms for the 4 models of persistenceof Argentine ants (Table 1). Vegetation cover and eleva-tion had a synergistic negative association with Argentineant persistence. The negative association between closedcanopy and persistence was mitigated with increasingwinter rainfall. Vegetation cover itself was not associatedwith Argentine ant appearance. However, for appearancefrom spring to autumn the positive association with thepresence of conspecific neighbors decreased as vegeta-tion cover increased.

As vegetation cover increased, appearance and persis-tence of winter ants increased. Logistic curves reachedhalf their maxima when the proportion of cover was0.53–0.77, depending on survey period (Figs. 3b & c).Vegetation cover was most consistently associated withappearance and persistence of winter ants and was se-lected first in 7 of the models (Table 2). The interactionbetween vegetation cover and Argentine ant presencewas selected in 3 of 8 models. The interaction strength-ened the association of winter ants with vegetation coverin the presence of Argentine ants and weakened it in theabsence of Argentine ants (Table 2 & Fig. 2b). Logistic

curves of the interaction terms reached half of their max-ima when the proportion of cover was 0.64–0.69 in thepresence of Argentine ants and 0.29–0.42 in their ab-sence. From spring or autumn to the next spring, vegeta-tion cover and proportion of conspecific neighbors werepositively and synergistically associated with persistenceof winter ants.

Elevation and Distance to Water

In general, appearance and persistence of Argentine antsincreased as elevation decreased (selected second to lastof 4, 4, and 8 variables in 3 of 8 models) (Table 1). Dis-tance to water, which was highly correlated with eleva-tion, was selected in only one model.

Similarly, the appearance and persistence of winterants were not associated consistently with proximity towater or elevation, and these variables were selected in1 of the 8 models. From autumn to spring, winter antspersisted more frequently at low-elevation sites, and theassociation between persistence and elevation increasedas distance to water increased.

Presence of Other Species

Sites with winter ants had a lower probability of Argen-tine ant appearance and persistence (selected fourth of8 and third of 7 terms in 2 of the 8 models). If only win-ter ants were observed at a site one autumn, Argentine



Tabl

e1.

Vari

able

sas

soci

ated

with

chan

ges

inth

eob

serv

eddi

stri

butio

nof

Arge

ntin

ean

tsin

Jasp

erRi

dge

Biol

ogic

alPr

eser

vein

nort

hern

Calif

orni

a(U

.S.A

.)af

ter

mod

else

lect

ion.

a

Bet

aco

effi

cien

tsof

com

mon

AIC

Sele

cted

model

cova

ria

tes

inse

lect

edm

odel

s

No.d

ata

poin

tsra

ndom

-A

rgen

tin

e(s

ites

,com

pa

red

effe

cts

sele

cted

log

no.

win

ter

vege

tati

on

an

tpre

sen

ceper

iods)

model

bm

odel

cli

kel

ihood

pa

ram

eter

sdin

terc

ept

nei

ghbors

edev

elopm

entf

vege

tati

on

gel

eva

tion

ha

nts

i:e

leva

tion

j

Ap

pea

ran

cesp

rin

gto

autu

mn

wit

hin

aye

ar

2193

(274

,17)

1158

896

−441

7−4

.35

0.71

(2)

−2.2

2(1

)–

–0.5

6(3

)–

–

autu

mn

of

year

xto

spri

ng

of

year

x+

1

1836

(238

,16)

772

621

−306

5−4

.91

0.49

(2)

−2.4

5(1

)–

––

–

spri

ng

of

year

xto

spri

ng

of

year

x+

1

1837

(256

,16)

938

706

−346

7−4

.57

0.94

(2)

−1.8

9(1

)–

–0.5

4(3

)–

–

autu

mn

of

year

xto

autu

mn

of

year

x+

1

1832

(233

,16)

1005

766

−372

11−3

.81

0.64

(2)

−2.0

2(1

)–

––0

.31

(4)

–

Per

sist

ence

spri

ng

toau

tum

nw

ith

ina

year

1724

(158

,17)

729

630

−305

104.

420.

77(1

)−0

.42

(3)

−0.7

9(2

)–

––

autu

mn

of

year

xto

spri

ng

of

year

x+

1

1837

(177

,16)

1340

1218

−601

82.

520.

65(2

)–0

.75

(1)

–0.4

0(3

)–

–0.

30(5

)

con

tin

ued



Tabl

e1

(con

tinue

d).

Bet

aco

effi

cien

tsof

com

mon

AIC

Sele

cted

model

cova

ria

tes

inse

lect

edm

odel

s

No.d

ata

poin

tsra

ndom

-A

rgen

tin

e(s

ites

,com

pa

red

effe

cts

sele

cted

log

no.

win

ter

vege

tati

on

an

tpre

sen

ceper

iods)

model

bm

odel

cli

kel

ihood

pa

ram

eter

sdin

terc

ept

nei

ghbors

edev

elopm

entf

vege

tati

on

gel

eva

tion

ha

nts

i:e

leva

tion

j

spri

ng

of

year

xto

spri

ng

of

year

x+

1

1608

(157

,16)

1070

949

−464

102.

810.

74(1

)−0

.36

(2)

−0.3

9(5

)–

–0.2

0(3

)0.

41(6

)

autu

mn

of

year

xto

autu

mn

of

year

x+

1

1842

(173

,16)

967

805

−390

123.

710.

90(1

)−0

.68

(2)

−0.6

8(3

)−0

.33

–0.

43(6

)

Arg

enti

ne

an

tsw

ere

obse

rved

du

rin

gth

ese

con

dbu

tn

ot

the

firs

tof

the

2su

rvey

per

iods.

aW

eu

sed

gen

era

lize

dli

nea

rm

ixed

model

sw

ith

logi

stic

resp

on

sea

nd

con

du

cted

forw

ard

sele

ctio

na

nd

ba

ckw

ard

elim

ina

tion

.We

stopped

addin

gva

ria

ble

sw

hen

no

new

vari

able

sre

du

ced

the

Aka

ike

info

rma

tion

crit

erio

n(A

IC)

by

more

tha

n2

;th

isre

sult

ing

model

wa

sth

ese

lect

edm

odel

.U

sin

gth

ispro

cedu

rew

ew

ere

able

tose

lect

asi

ngl

em

odel

for

each

tra

nsi

tion

.Lo

gli

kel

ihood,n

um

ber

of

pa

ram

eter

s,in

terc

ept,

an

dco

effi

cien

tsa

rere

port

edfo

rse

lect

edm

odel

son

ly.C

oef

fici

ents

wer

eca

lcu

late

dw

ith

sca

led

vari

able

s,a

nd

sca

lin

gva

ried

bet

wee

nco

mpa

riso

ns

of

dif

fere

nt

surv

eyper

iods;

thu

s,co

effi

cien

tssh

ou

ldbe

com

pa

red

on

lyw

ith

ina

sin

gle

tra

nsi

tion

.Nu

mber

sin

pa

ren

thes

esbes

ide

coef

fici

ents

indic

ate

ord

erin

wh

ich

term

sw

ere

sele

cted

.bIn

clu

ded

on

lyin

terc

ept

an

dra

ndom

effe

cts

of

site

an

dda

te.

c Sel

ecte

dw

ith

forw

ard

sele

ctio

nw

ith

ba

ckw

ard

elim

ina

tion

.dIn

terc

ept

an

d2

ran

dom

effe

cts

for

ever

yse

lect

edm

odel

an

dth

ese

lect

edva

ria

ble

sa

nd

inte

ract

ion

s.eP

roport

ion

of

surv

eyed

nei

ghbori

ng

site

sw

ith

Arg

enti

ne

an

tspre

sen

tdu

rin

gth

efi

rst

of

the

2su

rvey

per

iods.

fD

ista

nce

toth

en

eare

stdev

eloped

are

aw

ith

inth

epre

serv

eor

at

the

bou

nda

ry.

gP

roport

ion

of

the

site

cove

red

wit

hve

geta

tion

taller

tha

n0

.75

m.

hE

leva

tion

of

the

surv

eysi

te.

i Pre

sen

ce(1

)or

abse

nce

(0)

of

win

ter

an

tsdu

rin

gth

efi

rst

of

the

2su

rvey

per

iods.

j In

tera

ctio

nbet

wee

nve

geta

tion

cove

ra

nd

elev

ati

on

.



Tabl

e2.

Vari

able

sas

soci

ated

with

chan

ges

inth

eob

serv

eddi

stri

butio

nof

win

ter

ants

inJa

sper

Ridg

eBi

olog

ical

Pres

erve

inno

rthe

rnCa

lifor

nia

(U.S

.A.)

afte

rm

odel

sele

ctio

n.a

Sele

cted

Bet

aco

effi

cien

tsof

com

mon

AIC

model

cova

ria

tes

inse

lect

edm

odel

s

Win

ter

No.d

ata

poin

tsra

ndom

Arg

enti

ne

Arg

enti

ne

an

t(s

ites

,com

pa

red

effe

cts

log

no.

Arg

enti

ne

nei

gh-

an

ts:

an

ts:

tra

nsi

tion

per

iods)

model

bse

lect

edm

odel

cli

kel

ihood

pa

ram

eter

sdin

terc

ept

vege

tati

on

ea

nts

fbors

gdev

elopm

enth

vege

tati

on

i

Ap

pea

rin

gsp

rin

gto

autu

mn

wit

hin

aye

ar

2910

(302

,17)

1266

1173

−579

8−3

.50

0.94

(1)

–0.8

4(2

)–

0.55

(3)

0.34

(4)

autu

mn

of

year

xto

spri

ng

of

year

x+

1

3177

(293

,16)

2798

2655

−131

99

−1.6

40.

95(1

)−0

.057

(3)

0.16

(2)

0.65

(5)

–

spri

ng

of

year

xto

spri

ng

of

year

x+

1

2560

(288

,16)

2110

1973

−976

11−1

.73

0.90

(1)

−0.1

2(3

)0.

20(2

)0.

65(7

)–

autu

mn

of

year

xto

autu

mn

of

year

x+

1

3284

(292

,16)

1532

1454

−228

8−3

.14

0.90

(1)

−0.5

8(3

)–

0.47

(4)

0.50

(5)

Per

sist

ing

spri

ng

toau

tum

nw

ith

ina

year

988

(232

,15)

922

862

−421

10−1

.78

0.63

(1)

−0.3

9(2

)0.

47(6

)0.

41(5

)

autu

mn

of

year

xto

spri

ng

of

year

x+

1

367

(142

,16)

475

439

−210

100.

040.

66(1

)–

––

–

con

tin

ued



Tabl

e2

(con

tinue

d).

Sele

cted

Bet

aco

effi

cien

tsof

com

mon

AIC

model

cova

ria

tes

inse

lect

edm

odel

s

Win

ter

No.d

ata

poin

tsra

ndom

Arg

enti

ne

Arg

enti

ne

an

t(s

ites

,com

pa

red

effe

cts

log

no.

Arg

enti

ne

nei

gh-

an

ts:

an

ts:

tra

nsi

tion

per

iods)

model

bse

lect

edm

odel

cli

kel

ihood

pa

ram

eter

sdin

terc

ept

vege

tati

on

ea

nts

fbors

gdev

elopm

enth

vege

tati

on

i

spri

ng

of

year

xto

spri

ng

of

year

x+

1

802

(201

,15)

1044

973

−476

11−0

.09

0.60

(1)

–0.

26(3

)–

–

autu

mn

of

year

xto

autu

mn

of

year

x+

1

384

(141

,16)

474

459

−225

5−1

.05

––

––

–

Win

ter

an

tsw

ere

not

obse

rved

du

rin

gth

efi

rst

da

teof

the

“appea

rin

g”tr

an

siti

on

s.Th

eyw

ere

obse

rved

du

rin

gth

efi

rst

da

teof

the

“per

sist

ing”

tra

nsi

tion

s.aW

eu

sed

gen

era

lize

dli

nea

rm

ixed

model

sw

ith

logi

stic

resp

on

sea

nd

con

du

cted

forw

ard

sele

ctio

na

nd

ba

ckw

ard

elim

ina

tion

.We

stopped

addin

gva

ria

ble

sw

hen

no

new

vari

able

sre

du

ced

the

Aka

ike

info

rma

tion

crit

erio

n(A

IC)

by

more

tha

n2

;th

isre

sult

ing

model

wa

sth

ese

lect

edm

odel

.U

sin

gth

ispro

cedu

rew

ew

ere

able

tose

lect

asi

ngl

em

odel

for

each

tra

nsi

tion

.Lo

gli

kel

ihood,n

um

ber

of

pa

ram

eter

s,in

terc

ept,

an

dco

effi

cien

tsa

rere

port

edfo

rse

lect

edm

odel

son

ly.C

oef

fici

ents

wer

eca

lcu

late

dw

ith

sca

led

vari

able

s,a

nd

sca

lin

gva

ried

bet

wee

nco

mpa

riso

ns

of

dif

fere

nt

surv

eyper

iods;

thu

s,co

effi

cien

tssh

ou

ldbe

com

pa

red

on

lyw

ith

ina

sin

gle

tra

nsi

tion

.Nu

mber

sin

pa

ren

thes

esbes

ide

coef

fici

ents

indic

ate

ord

erin

wh

ich

term

sw

ere

sele

cted

.bM

odel

incl

udin

gon

lyin

terc

ept

an

dra

ndom

effe

cts

of

site

an

dda

te.

c Model

sele

cted

wit

hfo

rwa

rdse

lect

ion

wit

hba

ckw

ard

elim

ina

tion

.dP

ara

met

ers

cou

nte

din

clu

de

inte

rcep

ta

nd

2ra

ndom

effe

cts

for

ever

yse

lect

edm

odel

an

dth

ese

lect

edva

ria

ble

sa

nd

inte

ract

ion

s.eC

oef

fici

ent

for

the

pro

port

ion

of

the

site

cove

red

wit

hve

geta

tion

taller

tha

n0

.75

m.

fC

oef

fici

ent

for

the

pre

sen

ce(1

)or

abse

nce

(0)

of

Arg

enti

ne

an

tsdu

rin

gth

efi

rst

da

teof

the

tra

nsi

tion

.gC

oef

fici

ent

for

the

pro

port

ion

of

surv

eyed

nei

ghbori

ng

site

sw

ith

win

ter

an

tspre

sen

tdu

rin

gth

efi

rst

da

teof

the

tra

nsi

tion

.hC

oef

fici

ent

for

the

inte

ract

ion

bet

wee

nA

rgen

tin

ea

nt

pre

sen

cea

nd

the

dis

tan

ceto

the

nea

rest

dev

eloped

are

a.

i Coef

fici

ent

for

the

inte

ract

ion

bet

wee

nA

rgen

tin

ea

nt

pre

sen

cea

nd

vege

tati

on

cove

r.



Figure 2. (a) Relation between winter ant appearance from one spring survey to the next and distance of site todeveloped areas as a function of the presence of Argentine ants. Bars show the proportion of transitions from onespring to the next in which winter ants appeared during the second spring of the paired surveys (black bars, siteswith Argentine ants present during the first spring of the paired surveys; white bars, sites where Argentine antswere absent during the first spring of the paired surveys; near, 50% of sites closest to developed areas; far, 50% ofsites farthest from developed areas). (b) Relation between winter ant appearance from one spring survey to thenext and vegetation cover as a function of the presence of Argentine ants. Bars show the proportion of transitionsfrom one spring to the next in which winter ants appeared during the second spring of the paired surveys (black,sites with Argentine ants during the first spring of the paired surveys; white, sites without Argentine ants duringthe first spring; low and high, respectively, sites in the lowest and highest quartiles of proportion of vegetationcover with height >0.75 m).

ants were less likely to appear at that site the followingautumn (Table 1). Also, when both Argentine ants andwinter ants were observed at a site during spring, Argen-tine ants were less likely to be present at that site thefollowing spring (Table 1 & Supporting Information).

At sites where only Argentine ants were present, win-ter ants appeared less often in the next season or thenext year than at sites without Argentine ants (Table 2).At sites where winter ants and Argentine ants were bothobserved in the spring, winter ants persisted until the au-tumn less often than at sites where only winter ants wereobserved (Table 2). However, negative associations ofArgentine ants with appearance or persistence of winterants decreased as distance to development increased, asproportion of vegetation cover increased, and as winterrainfall decreased (Table 2 & Figs. 2a & b).

Weather

Argentine ant appearance and persistence between oneautumn and the next increased with increasing summerrainfall. This was the only weather variable that was as-sociated consistently with changes in Argentine ant pres-ence (selected third of 8 and fourth of 9 covariates, re-spectively, for models of appearance and persistence).The logistic curves reached half their maxima at 68 mm

of rainfall for appearance and 30 mm for persistence.Although temperature and its interaction with other vari-ables were selected for several models, their effects wereinconsistent and they were added late in the selectionprocess.

For winter ants, as for Argentine ants, associations ofappearance and persistence with weather variables wereequivocal. Neither temperature nor rainfall was associ-ated consistently with winter ant appearance or persis-tence.

Discussion

Human disturbance, habitat quality, and interactions withnative species may affect the distributions of invasivespecies and effects on native species in preserves neardeveloped areas. In our study, vegetation cover, ele-vation, presence of winter ants, and especially humandisturbance were all associated with changes in Argen-tine ant distribution. Vegetation cover and distance todeveloped areas also affected the association betweenArgentine ant presence and appearance by winter ants(Fig. 3).

Invasions usually begin in areas of human activity,largely because propagules are first introduced in theseareas (Sakai et al. 2001; Pysek et al. 2010; Rizali et al.



Figure 3. Seasonal patterns of association betweenvegetation cover with height >0.75 m and Argentineand winter ant appearance and persistence: effect ofvegetation cover on the probability (a) that Argentineants persisted at a site, (b) that winter ants appearedat a site, and (c) that winter ants persisted at a site(heavy solid lines, period from one autumn to thenext; light solid lines, from spring to autumn; heavydashed lines, from one spring to the next; light dashedlines, from autumn to spring). Probabilities werecalculated with the logistic functions computed forselected models (Table 1), and we assumed values ofall variables except vegetation cover were at theirmean.

2010). At regional and global scales, Argentine ants aremost successful in human-modified areas (Suarez et al.2001; Roura-Pascual et al. 2011). In our study, Argen-tine ants rarely invaded sites far from development, evenwhere they were present at neighboring sites. When theydid invade such a site, they did not persist (Fig. 1d). Ar-gentine ants’ association with development was not dueto opportunity only.

Rather, like other invasive species, including the fireant (Solenopsis invicta) (MacDougall & Turkington 2005;King & Tschinkel 2008), Argentine ants may succeed inthese sites because of human alterations to the physicalenvironment. The effects of distance to development atJasper Ridge may have been due to temperature rather

than water limitations. In arid southern California, greaterwater availability in and near suburban areas increases theprobability of invasions of Argentine ants (Suarez et al.1998; Holway & Suarez 2006; but see Bolger 2007). Inthe more mesic environment of Jasper Ridge, changesin distributions of Argentine ants were not associatedwith distance from water, although they did spread moreoften in rainy summers, which is consistent with find-ings of Heller et al. (2008). That Argentine ant appear-ance and persistence were lower at higher elevation mayhave been due to lower water availability (Ward 1987) orcooler temperatures at these sites (Hartley et al. 2010).Proximity to developed areas was associated with Argen-tine ant persistence especially after cold winters. ThatArgentine ant persistence was lower at sites shaded bytall vegetation suggests they may be temperature limitedat Jasper Ridge (Fig. 3a). Argentine ants do not occurin forested areas in Hawaii (Krushelnycky et al. 2005)and New Zealand (Ward & Harris 2005), probably due tocooler temperatures under the canopy. At Jasper Ridge,Argentine ants’ winter nest aggregations occur at sunnysites (Heller & Gordon 2006); possibly they cannot findsuitable winter nest sites in wooded areas, and migrateelsewhere or die during the winter.

Near Jasper Ridge, Argentine ants enter buildings dur-ing winter (Gordon et al. 2001), where warmer tempera-tures may lead to increased reproductive success (Hartley& Lester 2003; Abril et al. 2008). Argentine ants surviveand reproduce outdoors at Jasper Ridge, but may do soat rates too low to expand their distribution without mi-gration from surrounding developed areas. Bolger (2007)similarly proposed that the presence of Argentine antsnear suburban edges of southern California natural areasis due to a source–sink dynamic in which ants migratefrom high-quality, suburban habitat to low-quality less-developed habitat where they cannot sustain populationgrowth (Fagan et al. 1999).

Winter ant presence was associated with lower Argen-tine ant presence for some survey periods. Winter antpresence may impede Argentine ant invasion at highlyshaded sites that provide high-quality habitat for winterants but low-quality habitat for Argentine ants. Winterants were strongly associated with high levels of vege-tation cover (Table 2 & Figs. 3b & c), which is consis-tent with their distribution throughout North Americanwoodlands (Tschinkel 1987; Andersen 1997). In autumnsurveys, they were mostly limited to shaded sites, and Ar-gentine ants were less likely to appear the next autumnat these sites than at sites without winter ants. Presenceof winter ants at sites in the spring was also associatedwith lower persistence of Argentine ants (Supporting In-formation).

Similarly, appearance of winter ants at sites with Argen-tine ants varied with season, vegetation cover, and devel-opment. In most survey periods, winter ants persistedat sites regardless of Argentine ant presence or absence



(Table 2), possibly because, like several other ant speciesthat coexist with Argentine ants (Holway 1999; Holwayet al. 2002; Sagata & Lester 2009), they produce chem-icals to defend nests (Sorrells et al. 2011). Winter antappearance, however, was negatively associated with Ar-gentine ant presence in sites where Argentine ants weremost persistent: near development or without thick veg-etation cover (Figs. 2a & b).

Interactions with native species shape invasions mainlywhen the invading species is hindered by other factors(Levine et al. 2004; Ruesink 2007; Masciocchi et al. 2010).For ant invasions, in particular, resistance by native antsis usually ineffective (e.g., O’Dowd et al. 2003; Hoffmann& Saul 2010; reviewed in Lach & Thomas 2008); this istrue of Argentine ant invasions (Holway 1999; Carpinteroet al. 2007; Rowles & O’Dowd 2007). However, nativecompetitors may kill small laboratory colonies of Argen-tine ants (Walters & Mackay 2005; Sagata & Lester 2009),and native ant communities slow, although do not pre-vent, invasions of Argentine ants in Southern Californiaand Spain (Menke et al. 2007; Roura-Pascual et al. 2010).Sites in Europe with relatively many Leptomyrmecinegenera are less likely to be invaded by Argentine ants,but this effect is mainly apparent at sites at the edge ofthe Argentine ant’s climate tolerance where there is littlehuman disturbance (Roura-Pascual et al. 2011).

If buildings contribute to rapid population growth ofArgentine ants, then increased density of Argentine antsnear developed areas may lead to the exclusion of nativeants (Holway 1999; Walters & Mackay 2005; Sagata &Lester 2009). In some invasions, density reaches a crit-ical mass (Roura-Pascual et al. 2010) that facilitates awave of invasion (Von Holle & Simberloff 2005; Holle-bone & Hay 2007; Clark & Johnston 2009). At JasperRidge, in contrast, it appears that a flow of migrantsfrom developed areas has not helped Argentine antsovercome barriers to colonization (i.e., low-quality habi-tat and interaction with winter ants in the interior ofthe preserve). Thus, our results offer a local exampleof the regional pattern of Argentine ant invasion in Eu-rope (Roura-Pascual et al. 2011). The distribution of aninvasive species depends on habitat quality. At sites withlow habitat quality, interactions with native species mayprevent invasion, but human disturbance may changequality from low to high, and thereby alter the out-come of interactions between native species and invadingspecies.

Acknowledgments

J. Shors and the Jasper Ridge staff gave invaluable assis-tance and support for field work. Many Stanford under-graduates and community members have helped withant surveys over the years, particularly T. Riley in autumn2009. We thank P. Switzer for statistical advice, T. Hebert

for technical advice, P. Lester, and an anonymous re-viewer for advice on the manuscript, and S. Tuljapurkar,R. Dirzo, N. Chiariello, N. Heller, and J. Roughgarden formany helpful discussions.

Supporting Information

Calculations of values of environmental variables in Ar-cGIS (Appendix S1) and a figure showing the asso-ciation between presence of winter ants and Argen-tine ant persistence (Appendix S2) are available on-line. The authors are solely responsible for these ma-terials. Queries should be directed to the correspondingauthor.

Literature Cited

Abril, S., J. Oliveras, and C. Gomez. 2008. Effect of temperature onthe oviposition rate of Argentine ant queens (Linepithema humileMayr) under monogynous and polygynous experimental conditions.Journal of Insect Physiology 54:265–272.

Andersen, A. N. 1997. Functional groups and patterns of organizationin North American ant communities: a comparison with Australia.Journal of Biogeography 24:433–460.

Bates, D., and M. Maechler. 2009. lme4: Linear mixed-effects modelsusing S4 classes. R package version 0.999375-32. Available fromhttp://CRAN.R-project.org/package=lme4 (accessed April 2010).

Bini, L. M., et al. 2009. Coefficient shifts in geographical ecology: anempirical evaluation of spatial and non-spatial regression. Ecography32:193–204.

Bolger, D. T. 2007. Spatial and temporal variation in the Argentineant edge effect: implications for the mechanism of edge limitation.Biological Conservation 136:295–305.

Burnham, K. P., and D. R. Anderson. 1998. Model selection and infer-ence: a practical information-theoretic approach. Springer-Verlag,New York.

Carpintero, S., J. Reyes-Lopez, and L. Arias de Reyna. 2003. Impactof human dwellings on the distribution of the exotic Argentineant: a case study in the Donana National Park, Spain. BiologicalConservation 115:279–289.

Carpintero, S., J. Retana, X. Cerda, J. Reyes-Lopez, and L. Arias de Reyna.2007. Exploitative strategies of the invasive Argentine ant (Linep-ithema humile) and native ant species in a southern Spanish pineforest. Environmental Entomology 36:1100–1111.

Clark, G. R., and E. L. Johnston. 2009. Propagule pressure and distur-bance interact to overcome biotic resistance of marine invertebratecommunities. Oikos 118:1679–1686.

DiGirolamo, L. A., and L. R. Fox. 2006. The influence of abiotic factorsand temporal variation on local invasion patterns of the Argentineant (Linepithema humile). Biological Invasions 8:125–135.

Fagan, W. F., R. S. Cantrell, and C. Cosner. 1999. How habitat edgeschange species interactions. The American Naturalist 153:165–182.

Going, B. M., J. Hillerislambers, and J. M. Levine. 2009. Abiotic and bioticresistance to grass invasion in serpentine annual plant communities.Oecologia 159:839–847.

Gordon, D. M., L. Moses, M. Falkovitz-Halpern, and E. H. Wong. 2001.Effect of weather on infestation of buildings by the invasive Argen-tine ant, Linepithema humile (Hymenoptera: Formicidae). Ameri-can Midland Naturalist 146:321–328.

Hartley, S., and P. J. Lester. 2003. Temperature-dependent developmentof the Argentine ant, Linepithema humile (Mayr) (Hymenoptera:Formicidae): a degree-day model with implications for range limitsin New Zealand. New Zealand Entomologist 26:91–100.



Hartley, S., P. D. Krushelnycky, and P. J. Lester. 2010. Integrating phys-iology, population dynamics and climate to make multi-scale pre-dictions for the spread of an invasive insect: the Argentine ant atHaleakala National Park, Hawaii. Ecography 33:83–94.

Heller, N. E., and D. M. Gordon. 2006. Seasonal spatial dynamics andcauses of nest movement in colonies of the invasive Argentine ant(Linepithema humile). Ecological Entomology 31:499–510.

Heller, N. E., N. J. Sanders, J. W. Shors, and D. M. Gordon. 2008. Rainfallfacilitates the spread, and time alters the impact, of the invasiveArgentine ant. Oecologia 155:385–395.

Hoffmann, B. D. and Saul, W. C. 2010. Yellow crazy ant (Anoplolepisgracilipes) invasions within undisturbed mainland Australian habi-tats: no support for biotic resistance hypothesis. Biological Invasions12:3093–3108.

Hollebone, A. L., and M. E. Hay. 2007. Propagule pressure of an invasivecrab overwhelms native biotic resistance. Marine Ecology ProgressSeries 342:191—196.

Holway, D. A. 1999. Competitive mechanisms underlying the dis-placement of native ants by the invasive Argentine ant. Ecology80:238–251.

Holway, D. A. 2005. Edge effects of an invasive species across a naturalecological boundary. Biological Conservation 121:561–567.

Holway, D. A., and A. V. Suarez. 2006. Homogenization of ant commu-nities in mediterranean California: The effects of urbanization andinvasion. Biological Conservation 127:319–326.

Holway, D. A., L. Lach, A. V. Suarez, N. D. Tsutsui, and T. J. Case. 2002.The causes and consequences of ant invasions. Annual Review ofEcology and Systematics 33:181–233.

Human, K. G, S. Weiss, A. Weiss, B. Sandler, and D. M. Gordon. 1998.The effect of abiotic factors on the local distribution of the inva-sive Argentine ant (Linepithema humile) and native ant species.Environmental Entomology 27:822–833.

Ingram, K. K., and D. M. Gordon. 2003. Genetic analysis of dispersaldynamics in an invading population of Argentine ants, Linepithemahumile. Ecology 84:2832–2842.

Kestrup, A. M., and A. Ricciardi. 2009. Environmental heterogeneitylimits the local dominance of an invasive freshwater crustacean.Biological Invasions 11:2095–2105.

King, J. R., and W. R. Tschinkel. 2008. Experimental evidence thathuman impacts drive fire ant invasions and ecological change.Proceedings of the National Academy of Sciences 105:20339–20343.

Krushelnycky, P. D., S. M. Joe, C. C. Daehler, and L. L. Loope. 2005.The role of abiotic conditions in shaping the long-term patterns ofa high- elevation Argentine ant invasion. Diversity and Distributions11:319–331.

Lach, L., and M. L. Thomas. 2008. Invasive ants in Australia: documentedand potential ecological consequences. Australian Journal of Ento-mology 47:275–288.

Levine, J. M., P. B. Adler, and S. G. Yelenik. 2004. A meta-analysis of bi-otic resistance to exotic plant invasions. Ecology Letters 7:975–989.

MacDougall, A. S., and R. Turkington. 2005. Are invasive species thedrivers or passengers of change in degraded ecosystems? Ecology86:42–55.

Masciocchi, M., A. G. Farji-Brener, and P. Sackmann. 2010. Competitionfor food between the exotic wasp Vespula germanica and the na-tive ant assemblage of NW Patagonia: evidence of biotic resistance?Biological Invasions 12:625–631.

Menke, S. B., and D. A. Holway. 2006. Abiotic factors control invasionby Argentine ants at the community scale. Journal of Animal Ecology75:368–376.

Menke, S. B., R. N. Fisher, W. Jetz, and D. A. Holway, 2007. Bioticand abiotic controls of Argentine ant invasion success at local andlandscape scales. Ecology 88:3164–3173.

Nygard, J. P., N. J. Sanders, and E. F. Connor. 2008. The effects ofthe invasive Argentine ant (Linepithema humile) and the nativeant Prenolepis imparis on the structure of insect herbivore com-

munities on willow trees (Salix lasiolepis). Ecological Entomology33:789–795.

O’Dowd, D. J., P. T. Green, and P. S. Lake. 2003. Invasional ‘meltdown’on an oceanic island. Ecology Letters 6:812–817.

Pauchard, A., et al. 2009. Ain’t no mountain high enough: plant inva-sions reaching new elevations. Frontiers in Ecology and the Envi-ronment 7:479–486.

PRISM Climate Group. 2010. 4 km monthly grid data.PRISM Climate Group, Corvallis, Oregon. Available fromhttp://www.prism.oregonstate.edu/ (accessed April 2010)

Pysek, P., et al. 2010. Disentangling the role of environmental andhuman pressures on biological invasions across Europe. Proceedingsof the National Academy of Sciences 107:12157–12162.

Rizali, A., D. J. Lohman, D. Buchori, L. B. Prasetyo, H. Triwidodo, M. M.Bos, S. Yamane, and C. H. Schulze. 2010. Ant communities on smalltropical islands: effects of island size and isolation are obscured byhabitat disturbance and ‘tramp’ ant species. Journal of Biogeography37:229–236.

Roura-Pascual, N., J. M. Bas, and C. Hui. 2010. The spread of the Ar-gentine ant: environmental determinants and impacts on native antcommunities. Biological Invasions 12:2399–2412.

Roura-Pascual, N., et al. 2011. Relative roles of climatic suitability andanthropogenic influence in determining the pattern of spread in aglobal invader. Proceedings of the National Academy of Sciences108:220–225.

Rowles, A. D., and D. J. O’Dowd. 2007. Interference competition by Ar-gentine ants displaces native ants: implications for biotic resistanceto invasion. Biological Invasions 9:73–85.

Rowles, A. D., and J. Silverman. 2009. Carbohydrate supply lim-its invasion of natural communities by Argentine ants. Oecologia161:161–171.

Ruesink, J. L. 2007. Biotic resistance and facilitation of a non-nativeoyster on rocky shores. Marine Ecology Progress Series 331:1–9.

Sagata, K., and P. J. Lester. 2009. Behavioural plasticity associatedwith propagule size, resources, and the invasion success of theArgentine ant Linepithema humile. Journal of Applied Ecology 46:19–27.

Sakai, A. K., et al. 2001. The population biology of invasive species.Annual Review of Ecology and Systematics 32:305–332.

Sanders, N. J., K. E. Barton, and D. M. Gordon. 2001. Long-term dy-namics of the distribution of the invasive Argentine ant, Linep-ithema humile, and native ant taxa in northern California. Oecolo-gia 127:123–130.

Schilman, P. E., J. R. B. Lighton, and D. A. Holway. 2007. Water balancein the Argentine ant (Linepithema humile) compared with fivecommon native ant species from southern California. PhysiologicalEntomology 32:1–7.

Sorrells, T. R., L. Y. Kuritzky, P. G. Kauhanen, K. Fitzgerald, S. J.Sturgis, J. Chen, C. A. Dijamco, K. N. Basurto, and D. M. Gor-don. 2011. Chemical defense by the native winter ant (Prenolepisimparis) against the invasive Argentine ant (Linepithema hu-mile). Public Library of Science ONE 6:e18717. DOI: 10.1371/jour-nal.pone.0018717.

Suarez, A. V., D. A. Holway, and T. J. Case. 2001. Patterns of spreadin biological invasions dominated by long-distance jump dispersal:insights from Argentine ants. Proceedings of the National Academyof Sciences 98:1095–1100.

Suarez, A. V., D. T. Bolger, and T. J. Case. 1998. Effects of fragmen-tation and invasion on native ant communities in coastal SouthernCalifornia. Ecology 79:2041–2056.

Thomas, M. L., and D. A. Holway. 2005. Condition-specific competi-tion between invasive Argentine ants and Australian Iridomyrmex.Journal of Animal Ecology 74:532–542.

Tschinkel, W. R. 1987. Seasonal life history and nest architec-ture of a winter-active ant, Prenolepis imparis. Insectes Sociaux34:143–164.



Von Holle, B., and D. Simberloff. 2005. Ecological resistance to bi-ological invasion overwhelmed by propagule pressure. Ecology86:3212–3218.

Walters, A. C., and D. A. Mackay. 2005. Importance of large colony sizefor successful invasion by Argentine ants (Hymenoptera: Formici-dae): evidence for biotic resistance by native ants. Austral Ecology30:395–406.

Ward, D. F., and R. J. Harris. 2005. Invasibility of native habitats byArgentine ants, Linepithema humile, in New Zealand. New ZealandJournal of Ecology 29:215–219.

Ward, P. S. 1987. Distribution of the introduced Argentine ant(Iridomyrmex humilis) in natural habitats of the lower SacramentoValley and its effects on the indigenous ant fauna. Hilgardia 2:1–16.


effects of vegetation cover, presence of a native ant species, and human disturbance on colonization...

Documents